Transaction process data publishing method and system based on block chain and differential privacy

Abstract

The invention discloses a transaction process data publishing method and system based on a block chain and differential privacy, which is used for scenes with strict requirements on data privacy such as transaction amount and the like, and comprises the steps of sensitivity registration and privacy management; the transaction party initiates an analysis request and generates a task event; the under-chain computing node receives the task event and executes differential privacy noise addition calculation of transaction data in a trusted execution environment to generate a local proof result containing a data noise addition result, zero-knowledge proof and privacy budget; each computing node under the chain adopts a privacy protection aggregation proving protocol to recursively aggregate local proving results and construct a total proving; and performing on-chain verification on the aggregation result and the total proof, and issuing a verification event. In this way, the correctness, privacy and auditing performance of the query process are ensured, and the method is suitable for e-commerce, finance and other scenes with strict requirements for data privacy.

Description

Technical Field

[0001] This invention belongs to the technical field of privacy computing and blockchain, specifically relating to a method for publishing transaction process data based on blockchain and differential privacy. Background Technology

[0002] In a digital society where multi-party collaboration and data sharing are increasingly frequent, how to effectively publish sensitive transaction data while ensuring privacy has become one of the core research topics in the field of information security. This issue is particularly prominent in high-privacy scenarios such as e-commerce, finance, healthcare, and government affairs. On the one hand, the openness and sharing of data can significantly improve collaborative efficiency and business intelligence; on the other hand, the leakage of sensitive data may trigger serious privacy risks and compliance issues, especially in data access and analysis processes involving multiple parties.

[0003] Traditional data protection technologies such as access control and data encryption have mitigated the risk of data breaches to some extent, but they have inherent limitations in supporting flexible analysis, audit traceability, and verifiable computation.

[0004] In recent years, with the development of blockchain technology, decentralized ledger structures and verifiable smart contract mechanisms have provided a new approach to building transparent and trustworthy data exchange platforms. However, on-chain computation faces performance bottlenecks and cannot directly meet the needs of efficient privacy processing. Furthermore, differential privacy, as a statistical-level data protection technology, has been widely used to prevent sensitive information from being inferred during analysis, but ensuring the consistency of privacy budget configuration, perturbation strategies, and the verifiability of computation results in multi-node, multi-role collaborative environments remains a significant challenge.

[0005] At the same time, the development of zero-knowledge proof technology, especially the maturity of recursive proof and scalable proof systems, has provided a theoretical and engineering foundation for verifying computational correctness while maintaining data hiding.

[0006] Although existing research has attempted to apply the above technologies individually or in combination to data privacy protection systems, existing solutions often lack a unified architectural design and struggle to balance performance, privacy, and verifiability requirements. This is especially true when it is necessary to provide systematic guarantees on whether query behavior is compliant, whether the computation process is trustworthy, and whether the results are auditable.

[0007] Therefore, in order to address the privacy protection challenges in the current process of sharing transaction data among multiple parties, there is an urgent need for a systematic solution that integrates the verifiability of blockchain, the anonymity of differential privacy, and the computational auditability of zero-knowledge proofs, in order to achieve the goal of privacy release that ensures data does not leave the domain, privacy is controllable, and computation is verifiable. Summary of the Invention

[0008] In view of the above, the purpose of this invention is to provide a method for publishing transaction process data based on blockchain and differential privacy. It aims to construct a three-layer architecture system where rules are on-chain, computation is off-chain, verification is on-chain, and governance is configurable, fundamentally resolving the contradiction between data sharing and privacy protection.

[0009] To achieve the above-mentioned objectives, an embodiment provides a method for publishing transaction process data based on blockchain and differential privacy, comprising the following steps: Sensitive data registration and privacy governance; The transacting party initiates an analysis request and generates a task event; Off-chain computing nodes receive task events and perform differential privacy-preserving denoising computation on transaction process data in a trusted execution environment, generating a local proof result that includes the data denoising result, local zero-knowledge proof, and privacy budget. Each off-chain computing node recursively aggregates local proof results and constructs a total proof using a privacy-preserving aggregated proof protocol. The privacy-preserving aggregated proof protocol includes noisy result aggregation based on the number of samples, privacy budget aggregation that maintains the original privacy budget, and a total proof that simultaneously verifies the zero-knowledge proof and the noisy result aggregation results. Perform on-chain verification of the aggregation results and the total proof, and publish the verification event.

[0010] Preferably, sensitivity registration and privacy governance include: The operating platform or entrusted governance institution is responsible for registering legitimate data analysis functions and their sensitivity information on the smart contracts in the governance and verification layer. Specifically, this includes the mathematical expression of the data analysis function, the sensitivity S under the worst case and the derivation process, as well as the audit or confirmation documents matching the data analysis function. The audit documents are stored in a content-addressable decentralized storage system, and the unique hash identifier of the document is submitted to the smart contract.

[0011] Preferably, the transacting party initiates an analysis request and generates a task event, including: The requesting party initiates a data analysis request through a query contract on the governance and verification layer. The request includes a data analysis function identifier, a total privacy budget, and staked tokens. After the query contract verifies the legitimacy, it broadcasts a task event, providing an entry point for triggering off-chain tasks. The task event includes a query task identifier, a data analysis function identifier, and a total privacy budget.

[0012] Preferably, the off-chain computing node receives task events and performs differential privacy-preserving noise-adding computation on transaction data in a trusted execution environment, including: After receiving a task event, the off-chain computing node in the off-chain computing network reads encrypted data fragments from the basic data layer in a trusted execution environment, reconstructs the key and decrypts the data, analyzes the decrypted data using data analysis functions and sensitivity, and then executes a differential privacy mechanism to add noise to the processed data to generate a data noise-added result. The local zero-knowledge proof is used to verify the correctness of the computation process, including: the correctness of data noise addition operations performed based on fixed-point arithmetic, verifying that the noise data is sampled from a Laplace distribution that conforms to preset parameters, and verifying that the sensitivity used is consistent with the sensitivity registered on-chain. Preferably, each off-chain computing node recursively aggregates local proof results and constructs a total proof using a privacy-preserving aggregated proof protocol, including: Noise aggregation based on sample size: The noise aggregation result of each node is weighted and averaged according to the sample size. That is, the sample size of each node is used as the weight to multiply its own noise aggregation result, then the sum is calculated, and then the result is divided by the total sample size of all nodes to obtain the global noise aggregation result. Privacy budget aggregation that maintains the original privacy budget: The global privacy budget is the result of privacy budget aggregation; Simultaneously verify the zero-knowledge proof and the aggregation result of the noise-added result: verify the validity of the zero-knowledge proof of each node through the circuit, and also verify the correctness of the aggregation process of the noise-added result through the circuit, to obtain the total proof.

[0013] Preferably, when constructing the overall proof, the data of all nodes are aggregated from bottom to top according to the topological relationship of the nodes to construct a recursive proof tree.

[0014] Preferably, the aggregation result and the total proof are verified on-chain and a verification event is published, including: Any node submits the aggregation result and the total proof to the verification contract of the on-chain governance and verification layer for verification. The verification rules include: verifying the structural integrity of the total proof; checking the smart contract to determine whether the sensitivity S matches; reconstructing the aggregation path and verifying that the total privacy budget does not exceed the maximum privacy budget set by the system; and reviewing whether the privacy rules of each nested combination meet the differential privacy security boundary. Upon successful verification, a verification event is published, in which the verification documents include the task event ID and the global noise-adding result.

[0015] Preferably, the method further includes: implementing an exception handling and penalty mechanism, specifically including: if the verification fails or the community's questioning is valid, analyzing the failure type according to a two-layer failure model, wherein the first type is a technical failure and the second type is a trust challenge scenario; For technical failure types, a recovery window is provided for nodes experiencing technical failures, allowing them to submit proof of repair. For integrity challenge scenarios, community members are allowed to launch integrity challenges against suspicious nodes. When a challenge is successful, the challenger is rewarded, and the suspicious node is penalized with tokens and marked as untrustworthy.

[0016] Preferably, the off-chain node runs in a trusted execution environment supported by a remote verification mechanism, and the original data and key information are destroyed immediately after computation to ensure that the data is irrecoverable.

[0017] To achieve the above-mentioned objectives, embodiments of the present invention also provide a transaction process data publishing system based on blockchain and differential privacy, comprising: The basic data layer is used to store transaction process data; The governance and verification layer is used for sensitivity registration and privacy governance; it also supports transacting parties in initiating analysis requests and generating task events; and it is used for on-chain verification of aggregation results and total proofs and for publishing verification events. The off-chain computing network is used by off-chain computing nodes to receive task events and perform differential privacy-preserving denoising computation on transaction process data in a trusted execution environment, generating local proof results that include data denoising results, local zero-knowledge proofs, and privacy budgets. It is also used by each off-chain computing node to recursively aggregate local proof results and construct a total proof using a privacy-preserving aggregated proof protocol. The privacy-preserving aggregated proof protocol includes denoising result aggregation based on the number of samples, privacy budget aggregation that maintains the original privacy budget, and a total proof that simultaneously verifies the zero-knowledge proof and the denoising result aggregation results.

[0018] Compared with the prior art, the beneficial effects of the present invention include at least the following: 1. End-to-end trust minimization is achieved: By putting core rules (such as sensitivity) on the chain and enforcing them by ZKP, the system no longer relies on trust in off-chain computing nodes.

[0019] 2. Ensures the strength and auditability of privacy protection: The budget and implementation process of differential privacy become open and transparent, can be audited by anyone, and are guaranteed by the agreement not to be abused.

[0020] 3. Provides a scalable, high-performance solution: The original Privacy-Preserving Aggregate Proof Protocol (PPRP) effectively solves the performance bottleneck of distributed privacy computing, enabling it to be applied to large-scale datasets.

[0021] 4. Enhanced system economic security and robustness: The refined two-layer fault model effectively deters malicious behavior while protecting the interests of honest participants, ensuring the long-term healthy operation of the system.

[0022] Therefore, this invention provides a scalable and highly secure privacy-preserving data publishing solution, offering strong technical support for data-driven governance and compliance analysis applications. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the system logic layer of the transaction process data publishing method based on blockchain and differential privacy provided in the embodiment. Figure 2 This is a flowchart illustrating the transaction process data publishing method based on blockchain and differential privacy provided in the embodiment. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of this invention.

[0026] The inventive concept of this invention aims to resolve the structural conflict between privacy protection and trusted, verifiable computation in multi-party data collaboration. It proposes a method for publishing transaction process data based on blockchain and differential privacy, integrating on-chain governance, off-chain trusted computation, and a verifiable noise-adding mechanism. The transaction process data includes all data such as transaction amount, transaction time, number of transactions, logistics information, and credit score. This method is based on differential privacy theory and supported by blockchain and recursive zero-knowledge proofs, constructing a complete end-to-end system for data scheduling, computation execution, privacy control, and result verification.

[0027] like Figure 1 As shown, the core of this invention lies in constructing a hierarchical and collaborative computing architecture, which consists of three main parts: 1. Base Transaction Chain (L1) at the Foundation Data Layer: This layer serves as an immutable data storage medium for any blockchain network that supports smart contracts, such as Enterprise Ethereum, Hyperledger Fabric, or EVM-compatible public chains. Its primary responsibility is to immutably record raw business transaction and event data.

[0028] 2. Privacy Governance and Verification Chain (L2) at the Governance and Verification Layer: This layer is the core of privacy governance and the system control center. It mainly consists of a cluster of smart contracts deployed on L1 to solidify privacy rules on-chain. L2 does not directly store or process plaintext data. Its core functions include: registering sensitivity parameters, receiving data analysis requests, executing privacy protection policies, scheduling computational tasks, managing permissions and resources, and verifying and publishing the privacy-protected analysis results on-chain.

[0029] 3. Off-Chain Computing Network (OCN): This layer is a distributed network of nodes that perform computationally intensive tasks. It is responsible for reading the raw data (transaction data and / or event data) required from L1, performing complex privacy-preserving computation tasks (such as differential privacy processing and zero-knowledge proof generation) in a secure off-chain environment, ensuring the verifiability of the computation process through innovative proof protocols, and submitting the results and proofs to L2 for verification.

[0030] The entire computing architecture supports two implementation models: consortium blockchain and decentralized blockchain. The former is suitable for enterprise consortium scenarios, while the latter enables nodes to participate freely through economic incentive mechanisms.

[0031] like Figure 2 As shown in the embodiment, a method for publishing transaction process data based on blockchain and differential privacy includes the following steps: S1, Sensitivity Registration and Privacy Governance: The operating platform or entrusted governance organization is responsible for registering legitimate data analysis functions and their sensitivity information with the Registry.sol smart contract in the governance and verification layer (L2). Sensitivity is used to control the magnitude of noise added in the differential privacy mechanism, ensuring a controllable balance between accuracy and privacy, and avoiding privacy leaks caused by improper parameter settings.

[0032] Specifically, the registration process is implemented through the on-chain smart contract Registry.sol. The registration includes: the mathematical expression of the data analysis function f(D), the derivation of the worst-case sensitivity S, and an audit or authorization document matching the data analysis function. The registrant must store the audit document in the content-addressed decentralized storage system IPFS (InterPlanetary File System) and submit the document's unique hash identifier (CID) to the Registry.sol contract as an immutable on-chain credential.

[0033] For example, when the registered data analysis function is f(D) = average(D), that is, to calculate the average of the dataset D, and a single data cell x is set... i ∈ [0, M], the minimum size of the sample set is N minAccording to the definition of differential privacy, when two adjacent datasets D and D′ satisfy |DD′|1=1, the maximum difference (sensitivity) of their output functions is: S = max_{D,D′} |f(D) - f(D′)| = M / N min The sensitivity S mentioned above, and the analysis function identifier f id The IPFS hash of the audit documents will be uniformly registered in the smart contract Registry.sol.

[0034] Furthermore, the registration process also includes an on-chain governance approval mechanism to ensure that all adopted data analytics functions and sensitivities are compliant and trustworthy in both form and substance, specifically including: In consortium blockchain deployment scenarios: the registration of new data analysis functions requires approval by a multi-signature vote of more than 2 / 3 of the consortium members; In public network deployment scenarios: A "delayed activation + challenge period" mechanism is implemented. Each new registration proposal will enter a 7-day challenge window after submission. During this challenge window, any community member can stake tokens to launch a mathematical challenge, questioning the legality and accuracy of the sensitivity value calculation. If the challenge is successful, the registrant's stake will be forfeited and the reward will be given to the challenger; if there are no valid challenges, the registration will automatically enter the official activation state.

[0035] S2, the transaction party initiates an analysis request: the party requesting analysis submits a data analysis request to the query contract QueryRequest.sol, which includes parameters such as the data analysis function identifier, total privacy budget, and staked tokens. After the query contract verifies the legality, it broadcasts the query task event, providing an entry point for triggering off-chain tasks.

[0036] Specifically, stakeholders with analytical needs (such as businesses, regulatory agencies, or research organizations) can initiate specific data analysis requests through the L2 layer query contract QueryRequest.sol. The call parameters typically include: (1) QueryFunctionID, which represents the identifier of the data analysis function used in the request, such as “avg_delivery_time” or “sum_travel_cost”. This function has been pre-registered in the on-chain sensitivity smart contract Registry.sol; (2) ε total This represents the total privacy budget used in the request, which, under differential privacy mechanisms, represents the maximum privacy disclosure limit allowed by the task, and is usually a positive number (e.g., ε). total = 1.0), this value will determine the noise level that the system is allowed to perform during execution; (3) The amount of T tokens pledged is used for the economic constraint mechanism. This pledge is both an incentive guarantee for the requester's behavior and the basis for the system to call the penalty logic in the event of node misbehavior or request failure.

[0037] Upon receiving the call, the query contract QueryRequest.sol immediately verifies the legality and completeness of the aforementioned parameters, including checking whether QueryFunctionID has been registered, whether ε_total is within a reasonable range, and whether the token balance is sufficient and successfully locked. After successful verification, the contract generates a globally unique query task identifier QueryID and permanently writes the metadata (including the initiating address, function type, ε_total, timestamp, and staking information) into the on-chain state. Simultaneously, it publishes the event QueryDeclared(QueryID, QueryFunctionID, ε_total) as the event entry point for triggering off-chain tasks.

[0038] S3, Off-chain computing nodes receive tasks and process data: In the off-chain computing network, off-chain computing nodes listen for query task events, extract raw data from L1, compute the noisy result in the trusted execution environment, and construct a zero-knowledge proof that satisfies the differential privacy security boundary. The result is packaged into a local proof that includes the zero-knowledge proof, the noisy result, and the privacy budget.

[0039] Specifically, each node in the OCN listens for the QueryDeclared event on the chain, and initiates the corresponding computation process upon detecting a task. After the task is started, each OCN node extracts the required encrypted data fragment D from the basic data layer L1 according to the task description and data acquisition strategy. i It temporarily rebuilds the key, completes data decryption, computation, and privacy processing within the Trusted Execution Environment (TEE).

[0040] In this embodiment, OCN can be divided into a consortium blockchain version and a public network version. For the consortium blockchain version: OCN nodes are designated by consortium members, their identities are public and transparent, and data access permissions are jointly implemented through off-chain protocol negotiation and on-chain identity and permission management mechanisms, ensuring that data control is maintained among trusted organizations.

[0041] For the public network version: OCN nodes can be built by any PoS validator who has staked a large amount of tokens. To ensure the privacy of encrypted data and the security of computation, this invention proposes a short-term authorization protocol based on TEE and threshold decryption, used to authorize nodes to briefly access sensitive data and complete tasks: Since sensitive data on the L1 blockchain is stored encrypted with a public key, the corresponding private key is distributed to the OCN node network through (t,n) threshold sharding using threshold cryptography (such as the FROST protocol), where n is the total number of participating nodes, and t (t>n / 2) is the minimum number of nodes required to reconstruct the key. Only when the number of nodes that meet the threshold cooperate and collaboratively reconstruct the key in their TEEs can the data be decrypted. Before reconstructing the key, each node's TEE must provide remote proof (such as a Quote based on Intel SGX) to other participating nodes to verify the integrity and security of its operating environment, preventing the key from being reconstructed in an untrusted environment.

[0042] Furthermore, after the data is successfully decrypted, the actual closed-container computation phase begins. According to the Parallel Composition theorem for differential privacy, since each node performs computation on its independent local dataset D, the total privacy budget ε... total It can be distributed to each node without loss. The decrypted data is analyzed using data analysis functions and sensitivity metrics; specifically, each node first calculates its local average value μ. i = avg(D i Then, a differential privacy mechanism is executed to add noise to the processed data. This differential privacy mechanism is a Laplace mechanism, where the added noise is sampled from a Laplace distribution with a mean of 0 and a scale parameter equal to the ratio of sensitivity to privacy budget. Specifically, the Laplace noise η required for differential privacy is sampled. i Then, the node generates the noisy result r locally. i =μ i +η i And construct zero-knowledge proofs (ZKP) based on zk-SNARK or Bulletproof. i .

[0043] To address the compatibility issues between ZKP circuits operating in finite fields and floating-point calculations, all calculations (such as the local average value μ) are... i The sampling noise η i Sensitivity S, Privacy Budget ε total Noise added result r i All are discretized using fixed-point arithmetic, i.e., multiplied by a large scaling factor (such as 2). 64 And round down. This ZKP is used to prove the satisfaction of the following circuit constraints: 1) DP satisfaction: r i =μ i +η i The circuit verifies that the addition relationship holds true under fixed-point conditions.

[0044] 2) Noise distribution: Verify η i It is sampled from a correctly parameterized Laplace distribution. This is achieved through a sub-circuit that samples η. i The validity of its source is demonstrated with high confidence by comparing it with a set of pre-computed probability ranges representing the target distribution.

[0045] 3) Sensitivity Matching: The circuit contains a logic that accesses a read-only copy of an on-chain state (or provided by an oracle) via QueryFunctionID to verify that the sensitivity S used is exactly the same as the value registered in the on-chain Registry.sol contract.

[0046] Finally, each OCN node outputs a tuple of local proof results. i= <r i ZKP i ε total The original data and reversible information are cleared in the TEE in real time to avoid the risk of leakage and reproduction.

[0047] S4, Recursive Aggregation and Construction of the Overall Proof: The local proof results of each node are recursively aggregated, and the noisy results of multiple nodes are uniformly aggregated, the privacy budget is accumulated, and the cross-node differential privacy proof is recursively encapsulated to construct a recursive proof tree. Finally, the aggregated result and the unified verification of the global overall proof result are generated. The aggregated result includes the global noisy result and the overall privacy budget.

[0048] This invention proposes a core innovative mechanism—the Privacy-Preserving Recursive Proof Protocol (PPRP)—combined with the AggregationCircuit structure. This protocol and circuit structure together constitute the key technical link in solving the problem of trusted aggregation and verification of multi-source privacy computation results, achieving recursive aggregation and the construction of the overall proof.

[0049] Specifically, the noise addition result aggregation based on the number of samples is as follows: the noise addition result of each node is weighted and averaged according to the number of samples. That is, the number of samples of each node is used as the weight to multiply its own noise addition result, then the sum is calculated, and then the result is divided by the total number of samples of all nodes to obtain the global noise addition result. Privacy budget aggregation that maintains the original privacy budget: The global privacy budget is the result of privacy budget aggregation; Simultaneously verify the zero-knowledge proof and the aggregation result of the noise-added result: verify the validity of the zero-knowledge proof of each node through the circuit, and also verify the correctness of the aggregation process of the noise-added result through the circuit, to obtain the total proof.

[0050] More specifically, for any two local proof results tuple_A and tuple_B, their corresponding aggregation logic is as follows: Noise addition result aggregation: Weighted average of the noise addition results by sample size: rAB=(nA⋅rA+nB⋅rB) / (nA+nB), where nA and nB represent the sample size of nodes A and B respectively, and rA and rB represent the noise addition results of nodes A and B respectively.

[0051] Privacy budget aggregation: Since the nodes are combined in parallel, the global privacy cost is the maximum value calculated independently by each node. Therefore, the aggregated privacy budget ε global Still ε total No need to accumulate.

[0052] Zero-knowledge proof aggregation: The aggregation circuit receives two zero-knowledge proofs (ZKPs). A and ZKP B As input, these are verified to be valid. Subsequently, the circuit verifies the correctness of the weighted average calculation of rAB. Upon successful verification, a new aggregated global proof (ZKP) is generated. AB The validity of all its subcomputations was recursively verified.

[0053] The aggregation process constructs a recursive proof tree from bottom to top according to the topological relationships of the nodes, and the system finally generates a globally noisy average value r. global This is accompanied by a globally verified proof. This proof structure not only verifies each step of the local differential privacy processing but also effectively encapsulates the mathematical correctness of all aggregation intermediate processes.

[0054] S5, On-chain Verification and Result Publication: Submit the aggregated results and total proof to the on-chain verification contract to verify its structural correctness and differential privacy compliance. After passing the verification, a VerifiedEvent is published and the results are stored on-chain. Stakers can retrieve their tokens.

[0055] After completing the overall proof construction, the next step is to proceed with the on-chain verification process to ensure that the final output has publicly verifiable and trustworthy characteristics on the blockchain. Specifically, the system allows any participant to submit the quadruple ⟨QueryID, r global , ε total,globalproof> Submit to the on-chain verification contract for inspection.

[0056] Specifically, any participant can use ⟨QueryID, r global , ε total The `proof_global` method submits a verification contract to the on-chain governance and verification layer. The contract verification rules include: verifying the integrity of the `proof_global` structure; checking the `Registry.sol` smart contract to determine if the sensitivity S matches; and reconstructing the aggregation path and verifying ε. total Not exceeding the system setting ε max And verify whether the privacy rules of each nested combination meet the differential privacy security boundary.

[0057] If verification is successful, the VerifiedEvent(QueryID, r) event will be triggered. global The results are broadcast and the analysis output is written to on-chain permanent storage. The analysis requester can retrieve their staked tokens T after the unlocking period. If verification fails, or during the arbitration period τ... fix If other nodes or third-party participants submit valid questions or reports regarding this task, the system will automatically enter the exception handling process for adjudication.

[0058] S6, Anomaly Handling and Penalty Mechanism: If verification fails or community objections are valid, the system analyzes the fault type based on a two-layer fault model and executes mechanisms such as recovery cycle, penalty deduction, and arbitration ruling to ensure the credibility, fairness, and reliability of the system.

[0059] To ensure the stable operation of the system and the authenticity and reliability of each analysis result before it is published on the blockchain, this invention further introduces a two-layer fault model to identify different types of faults in off-chain nodes during the computation and submission process, and implements differentiated response strategies accordingly.

[0060] The system distinguishes between data format errors (technical faults) and logical / parameter tampering (integrity issues) by verifying the specific error codes of the Proof verification failure generated by ZKP.

[0061] The first category is technical failures, where a node fails verification due to non-malicious factors such as calculation errors, missing parameters, or system anomalies. Such nodes will not be immediately penalized; their staked tokens (T) will be temporarily locked, with a recovery window (τ) provided. recover During this period, nodes can resubmit the corrected results and proofs. If the verification passes, the staked tokens will be automatically unlocked and no violation will be recorded. If the repair is not completed within the time limit, it will be considered as insufficient capability, and the system will evaluate and restrict it in the next governance cycle.

[0062] The second category is the integrity challenge scenario, targeting nodes that, while passing formal verification, actually engage in malicious behavior (such as forging input or using illegal sensitivity parameters). The system allows other nodes or participants to initiate challenges, accompanied by verifiable on-chain evidence. Upon receiving a challenge, the verification contract adjudicates it through voting, arbitration, or rule-based judgment. If the challenge is successful, the node will be identified as a malicious actor, and the system will immediately implement penalties: including confiscating all staked tokens, distributing most of the reward to the challenger, depositing the remainder into the system's public treasury, recording its negative reputation, and restricting or permanently banning it from participating again. This mechanism combines community oversight and economic incentives to ensure the system operates reliably, fairly, and sustainably.

[0063] The invention innovatively introduces a combination strategy of sensitivity governance, differential privacy mechanism, zero-knowledge proof, and blockchain notarization. While protecting the privacy of data sources, it enables the credible aggregation and verifiable publication of results from multiple nodes, providing a new system paradigm for privacy-compliant data analysis. It is widely applicable to various scenarios such as financial regulatory data release, supply chain visualization, and medical and scientific research data analysis.

[0064] The specific embodiments described above illustrate the technical solution and beneficial effects of the present invention in detail. It should be understood that the above description is only the most preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, additions, and equivalent substitutions made within the scope of the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for publishing transaction process data based on blockchain and differential privacy, characterized in that, Includes the following steps: Sensitive data registration and privacy governance; The transacting party initiates an analysis request and generates a task event; Off-chain computing nodes receive task events and perform differential privacy-preserving denoising computation on transaction process data in a trusted execution environment, generating a local proof result that includes the data denoising result, local zero-knowledge proof, and privacy budget. Each off-chain computing node recursively aggregates local proof results and constructs a total proof using a privacy-preserving aggregated proof protocol. The privacy-preserving aggregated proof protocol includes noisy result aggregation based on the number of samples, privacy budget aggregation that maintains the original privacy budget, and a total proof that simultaneously verifies the zero-knowledge proof and the noisy result aggregation results. Perform on-chain verification of the aggregation results and the total proof, and publish the verification event.

2. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, Sensitive registration and privacy governance, including: The operating platform or entrusted governance institution is responsible for registering legitimate data analysis functions and their sensitivity information on the smart contracts in the governance and verification layer. Specifically, this includes the mathematical expression of the data analysis function, the sensitivity S under the worst case and the derivation process, as well as the audit or confirmation documents matching the data analysis function. The audit documents are stored in a content-addressable decentralized storage system, and the unique hash identifier of the document is submitted to the smart contract.

3. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, The transacting party initiates an analysis request and generates a task event, including: The requesting party initiates a data analysis request through a query contract on the governance and verification layer. The request includes a data analysis function identifier, a total privacy budget, and staked tokens. After the query contract verifies the legitimacy, it broadcasts a task event, providing an entry point for triggering off-chain tasks. The task event includes a query task identifier, a data analysis function identifier, and a total privacy budget.

4. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, Off-chain computing nodes receive task events and perform differential privacy-preserving noise-adding computation on transaction data in a trusted execution environment, including: After receiving a task event, the off-chain computing nodes in the off-chain computing network read non-overlapping encrypted data fragments from the basic data layer in a trusted execution environment, reconstruct the key and decrypt the data, analyze the decrypted data using data analysis functions and sensitivity, and then execute a differential privacy mechanism to add noise to the processed data to generate a data-noised result. The local zero-knowledge proof is used to verify the correctness of the computation process, including: the validity of the data noise addition process, the noise being sampled on a correctly parameterized Laplace distribution, and the sensitivity used being consistent with the registered sensitivity.

5. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, Each off-chain compute node recursively aggregates local proof results and constructs a total proof using a privacy-preserving aggregated proof protocol, including: Noise aggregation based on sample size: The noise aggregation result of each node is weighted and averaged according to the sample size. That is, the sample size of each node is used as the weight to multiply its own noise aggregation result, then the sum is calculated, and then the result is divided by the total sample size of all nodes to obtain the global noise aggregation result. Privacy budget aggregation that maintains the original privacy budget: The global privacy budget is the result of privacy budget aggregation; Simultaneously verify the zero-knowledge proof and the aggregation result of the noise-added result: verify the validity of the zero-knowledge proof of each node through the circuit, and also verify the correctness of the aggregation process of the noise-added result through the circuit, to obtain the total proof.

6. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1 or 5, characterized in that, When constructing the overall proof, the data of all nodes are aggregated from bottom to top according to the topological relationship of the nodes to build a recursive proof tree.

7. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, Perform on-chain verification of the aggregation results and the total proof, and publish verification events, including: Any node submits the aggregation result and the total proof to the verification contract of the on-chain governance and verification layer for verification. The verification rules include: verifying the structural integrity of the total proof; checking the smart contract to determine whether the sensitivity S matches; reconstructing the aggregation path and verifying that the total privacy budget does not exceed the maximum privacy budget set by the system; and reviewing whether the privacy rules of each nested combination meet the differential privacy security boundary. Upon successful verification, a verification event is published, in which the verification documents include the task event ID and the global noise-adding result.

8. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, The method also includes: implementing an exception handling and penalty mechanism, specifically including: if verification fails or community questioning is valid, analyzing the failure type according to a two-layer failure model, where the first type is a technical failure and the second type is a trust challenge scenario; For technical failure types, a recovery window is provided for nodes experiencing technical failures, allowing them to submit proof of repair. For integrity challenge scenarios, community members are allowed to launch integrity challenges against suspicious nodes. When a challenge is successful, the challenger is rewarded, and the suspicious node is penalized with tokens and marked as untrustworthy.

9. The method for publishing transaction process data based on blockchain and differential privacy according to claim 1, characterized in that, Off-chain nodes run in a trusted execution environment supported by remote verification mechanisms. The original data and key information are destroyed immediately after computation, ensuring that the data is irrecoverable.

10. A transaction process data publishing system based on blockchain and differential privacy, characterized in that, include: The basic data layer is used to store transaction process data; The governance and verification layer is used for sensitivity registration and privacy governance. It is also used to support trading parties in initiating analysis requests and generating task events; it is also used to perform on-chain verification of aggregation results and total proofs and publish verification events; The off-chain computing network is used by off-chain computing nodes to receive task events and perform differential privacy-preserving denoising computation on transaction process data in a trusted execution environment, generating local proof results that include data denoising results, local zero-knowledge proofs, and privacy budgets. It is also used by each off-chain computing node to recursively aggregate local proof results and construct a total proof using a privacy-preserving aggregated proof protocol. The privacy-preserving aggregated proof protocol includes denoising result aggregation based on the number of samples, privacy budget aggregation that maintains the original privacy budget, and a total proof that simultaneously verifies the zero-knowledge proof and the denoising result aggregation results.

Images